Využití služeb IBM Watson pro porozumění textu

Bachelors thesis deals with artificial intelligence system from IBM (International Business Machines) company named Watson. In the theoretical part there are described all offered functionalities of the system at the time of writing, basic principles of functioning, history and current usage. This part of the work fulfills the first goal of becoming familiar with IBM Watson and its framework. The practical part is about using the system for selected text comprehension tasks. During the creation of this part of the work, two objectives were met, namely getting to know the area of understanding text in general and with using IBM Watson services and to test IBM Watson's understanding of text. The system is mainly evaluated based on reference data sets previously used for machine learning in other systems. Three functionalities were selected for the evaluation, namely the service for the recognition and classification of entities, sentiments and categories. The system outputs were compared with the reference values in the data sets and based on this were calculated, among other things, precision and recall, which indicates the success of the system. These values are also listed for Google's competitive system for sentiment and category recognition features. At the end of the work, the last goal is met, and IBM Watson's services are compared to the services available from Google. Here the results are evaluated both in comparison with the reference data sets and with the competitive system.